Why the real risk to enterprise infrastructure isn’t technology, it’s talent.
The mainframe is not in crisis. It processes billions of transactions every day with the same quiet reliability it has delivered for decades. The banking system runs on it. Insurance claims flow through it. Healthcare records depend on it. As a platform, it has outlasted every prediction of its obsolescence and earned its place at the center of enterprise computing.
The crisis, when it arrives, will not come from a technology failure. It will come from a workforce failure.
Mainframe professionals have always occupied a unique position inside their organizations. They understand not just how the systems work, but how the business works, down to the transactional level. They know where the data lives, how it moves, and what the warning signs look like before something breaks. That depth of knowledge is extraordinary, and it comes with a predictable consequence: when you are the most capable person in the room, you receive the most work.
What begins as a reasonable extension of responsibilities has, in many organizations, become an unsustainable accumulation of them. A mainframe developer fluent in COBOL and JCL is asked to adopt Python scripting. Then to integrate CI/CD pipeline tooling. Then, after a C-suite executive encounters an article about artificial intelligence, to evaluate large language models, build AI governance frameworks, and explain inferencing to a board of directors.
Each of these requests, considered in isolation, is defensible, but when they are evaluated together, they depict a mainframe professional who is simultaneously expected to function as a DevOps engineer, data architect, security analyst, and AI strategist while still ensuring the uninterrupted operation of the most critical infrastructure in the enterprise. Expectations expand, but headcount does not.
The pressure is most acute at organizations where the mainframe team is small. A shop running dozens of LPARs may have only three or four people who truly understand the environment from end to end. Those individuals become the bottleneck for every initiative that touches the platform, and in large enterprises, nearly every initiative does. The system of record for banking, insurance, healthcare, and manufacturing runs on the mainframe. No modernization effort proceeds without the involvement of the people who know it best.
When organizations encounter this constraint, a tempting response is to propose wholesale replacement of the platform itself, a strategy that mistakes the source of the problem. Replacing foundational infrastructure to solve a talent shortage is the organizational equivalent of demolishing a house because you cannot find an experienced plumber. The platform is not the liability. The failure to invest in the people who operate it is.
Many of the most experienced mainframe professionals in the workforce today have been in the field for 30 to 45 years. They have carried organizations through Y2K, through the client-server era, through the rise of cloud computing, and through successive waves of predictions that the mainframe would soon be obsolete. Their institutional knowledge is not merely technical; it is historical and therefore critical. They understand why a particular batch job runs at two in the morning on the third Tuesday of the month. They know which internal department transmits a malformed data set every March and have written the compensating logic to handle it.
Thousands of these professionals are approaching retirement, and the pipeline of replacements is thin. New talent is entering the field, sometimes by deliberate recruitment, more often by accident, as developers hired for other roles discover a mainframe team that urgently needs them. Serendipity is not a workforce strategy. And the distance between “recently learned COBOL” and “capable of managing 175 LPARs at a major insurer” is not bridged by a training course. It is measured in years of accumulated, context-specific knowledge that has never been written down.
Acknowledging the problem is not sufficient. Proactive intentional actions aligned to a broader business resiliency strategy are how organizations must approach this issue.
Stop treating the mainframe team as a general-purpose resource. Adopting modern tooling, such as Git on z/OS, CI/CD integration, cloud connectivity, etc., represents genuine value for these teams, and they should be equipped to pursue it. But that transition requires dedicated time, training, and resourcing. Adding it to an already overloaded schedule does not produce modernization. It produces burnout.
Invest in structured mentorship before it is too late. The transfer of institutional knowledge from veteran professionals to their successors cannot happen through documentation alone. It requires sustained, intentional pairing of experienced and emerging talent. Organizations without succession plans are not merely at risk of a skills gap; they are at risk of losing knowledge that jeopardizes system performance.
Apply honest scrutiny to AI mandates. Artificial intelligence has genuine potential in this space, particularly for large-scale data analysis and productivity tooling. But executives who encounter an AI headline in a magazine stuffed in the seat-back pocket of an airline seat should not be setting the technical agenda for teams responsible for the most critical infrastructure in the enterprise. Leadership mandates issued without platform fluency creates anxiety and busywork – the antithesis of innovation.
Treat retention as a strategic priority. The professionals who operate the mainframe are not interchangeable staff filling a functional role. They are the institutional memory of your business – the custodians of decades of decisions, workarounds, and hard-won operational wisdom. The conditions that cause them to leave, and the knowledge they take with them when they do, are not recoverable on any reasonable timeline.
Innovation on the mainframe will continue long into the future, and many mainframe customers will continue to depend upon its capabilities for generations to come. It will process transactions, run batch jobs, and anchor enterprise systems long after this moment has passed. The technology has proven its resilience repeatedly and comprehensively.
So the question is not whether the platform will endure. The question is whether the organizations that depend on it will have made the investments in people, in succession planning, and in honest resource allocation to ensure that their teams remain sharp in their efforts to support this amazing platform 24 hours every day, 365 days every year.